Does higher Institutional Quality improve the Appropriateness of Healthcare Provision?

Highlights

• Institutional quality affects the appropriateness of healthcare provision.

• Grey areas of medicine, subject to doctor discretion, are more likely to be affected.

• A standard deviation increase in institutional quality reduces C-sections by 10 pp.

Abstract

We study the effect of institutional quality on the appropriateness of healthcare provision in Italian hospitals. We focus on cesarean section rates for first-time mothers, which is a common indicator of appropriateness in healthcare and is vulnerable to providers’ opportunistic behaviors. To identify the causal effect of institutional quality we rely on an IV strategy based on historical instruments, exploiting the idea that current differences in institutional quality across regions have been shaped by different cultural and political histories. We find that a standard deviation increase in our indicator of institutional quality leads to a decrease of about 10 percentage points in cesarean section rates. Our results are robust to different measures of institutional quality and samples.

Introduction

Institutions matter. From the seminal work of North, 1981, North, 1990, North, 1991, the role of institutions in affecting behaviors in society has been widely recognized. Institutions influence economic performance through various mechanisms.¹ The institutional framework affects investment and growth (Mauro, 1995), the effectiveness of foreign aid and the expected benefits of natural resource endowments (Burnside and Dollar, 2000, Mehlum et al., 2006), firms’ efficiency (Dal Bó and Rossi, 2007), and even the emergence of organized crime (Acemoglu et al., 2020).

In the public sector, where public officials need to be kept accountable with implicit incentives (e.g., Alesina and Tabellini, 2007, Alesina and Tabellini, 2008), the role of the institutional framework is of primary importance. Mauro (1998) studies the composition of public expenditure across over 100 countries, and finds that countries where the perception of corruption is higher spend less on sectors which provide less lucrative opportunities for public officials. Similarly, Gupta et al. (2001) report that corruption is associated with higher military spending, suggesting the presence of rent-seeking behavior. In the political sector, Nannicini et al. (2013) show that the electoral punishment of politicians’ misbehaviors (such as absenteeism in Italian Parliament) is considerably larger in districts with higher social capital. Wong et al. (2017) examine the effects of an institutional reform implemented in China, and find that a higher quality of governance in rural villages increases the quality of public infrastructure provision.

In this study we investigate the effect of institutional quality on the appropriateness of healthcare provision in Italian hospitals. Healthcare spending in OECD countries represents about 9% of GDP and the majority of health expenditure is publicly financed (OECD, 2017). Healthcare accounts for about 15% of total public spending, though in some countries such as the United States and Germany more than 20% of public spending is for healthcare (OECD, 2017). Health spending is expected to grow further driven by an ageing population and technological innovation, raising challenges to the sustainability of public finances. Increasing the appropriateness in healthcare provision is thus critical in making health systems sustainable.

Due to the asymmetry of information which characterizes the doctor-patient relationship (Arrow, 1963), healthcare is vulnerable to providers’ opportunistic behaviors, which may lead to under- or over-treatment (Ellis and McGuire, 1986). This is also the case within the hospital sector, where providers respond to financial incentives across a range of health systems, including those based on a National Health Service where hospital physicians are typically salaried (see Section 3 for more detailed discussion). The asymmetry of information is further exacerbated by the difficulty for public insurers to measure providers’ performance, which further reduces their accountability (European Commission, 2017).

We focus on cesarean section rates during childbirth in Italy as a measure of appropriateness of healthcare provision. Cesarean section rates for first-time mothers are recognized as a valid indicator of appropriateness in the provision of healthcare services both in the literature and by policymakers (Baicker et al., 2006, OECD, 2009). In absence of clinical reasons or complications (e.g., uterine rupture), vaginal delivery is the recommended mode of delivery. Cesarean section is more invasive, it involves risks during surgery, has longer recovery times following birth and is more costly. For example, cesarean delivery has been found to increase the risk of maternal mortality and re-hospitalization, infant morbidity, and is negatively associated with child cognitive development (Lydon-Rochelle et al., 2000, Villar et al., 2006, Li et al., 2014, Polidano et al., 2017).

Crucially, the decision making in childbirth falls in the gray area of medicine, implying that it is possible for physicians to argue that a cesarean section is appropriate when it could be avoided (Chandra et al., 2011, Johnson and Rehavi, 2016). This makes the choice of childbirth delivery particularly exposed to the discretion of physicians (Gruber and Owings, 1996, Hopkins, 2000, Di Giacomo et al., 2017) and to what in the health economics literature is known more broadly as “supplier-induced demand” (Rice, 1983).²

There are several mechanisms through which local institutional quality can affect the appropriateness of hospital services. Regional authorities can measure and monitor caesarean sections for each hospital, publish them in the public domain to “shame” providers with excessive rates or to trigger audits and ask for detailed accounts of their treatment choices. They can also reduce the price differential between a cesarean section and a natural delivery, to dampen or eliminate the financial incentives. The extent to which these measures are introduced will depend on the quality of local administration, their accounting systems, and their determination and culture to reduce waste of public resources. Aside from formal measures, a local institutional environment characterized by widespread corruption, poor governance, and weak rule of law can reinforce hospital incentives to over-treat.

Italy exhibits large variations in cesarean section rates across regions that cannot be explained by sociodemographic and clinical factors (e.g., Francese et al., 2014, Guccio and Lisi, 2016). In turn, this has led to concerns by the Italian Ministry of Health that regularly monitors cesarean section rates across regions and providers (Ministry of Health, 2017). The cesarean section rate in Italy was 35% in 2016 and varied between 20% in Trentino Alto Adige and 59% in Campania, well beyond the level deemed appropriate by the World Health Organization (1985) of around 20%, as for instance observed in Nordic countries (OECD, 2017).

We measure institutional quality in Italian regions with the Institutional Quality Index, a multidimensional quantitative indicator (Nifo and Vecchione, 2014) based on the hierarchy framework employed by the World Bank Worldwide Governance Indicators (Kaufmann et al., 2010). As an alternative measure, we also employ a perception-based multidimensional indicator, the European Quality of Government Index (EQI) (Charron et al., 2014).³ Indicators of the quality of institutional environment (e.g., corruption, quality of government) display large differences across Italian regions (Golden and Picci, 2005, Charron et al., 2014). For example, Charron et al. (2014, p. 74) notice that “the gap between Bolzano and Campania in the data is much larger than the gap in the national averages between Denmark and Portugal”. Such regional variation makes Italy a particularly suitable case study to investigate the effect of institutional quality (e.g., Nannicini et al., 2013, Castro et al., 2014, Lasagni et al., 2015).

To identify the causal effect of institutional quality on the appropriateness of healthcare services, we employ an instrumental variable approach based on historical data. We follow the influential literature which argues that current institutional backwardness is also a byproduct of history (e.g., North, 1981, Acemoglu et al., 2001, Tabellini, 2010, Guiso et al., 2016): current differences in institutional quality across Italian regions have been shaped by different cultural and political histories. Specifically, our instruments are an indicator of quality of political institutions in the period from 1600 to 1850 and the percentage of population over age 6 able to read in 1881, provided in Tabellini (2010).

Our results show that institutional quality improves the appropriateness in the provision of childbirth services. A standard deviation increase in our indicator of institutional quality leads to a reduction of about 10 percentage points in cesarean section rates. The results are robust to alternative indicators of institutional quality and samples. We also find that, amongst the different dimensions of institutional quality, corruption and government effectiveness in the local area appear to be the most relevant in affecting the provision of childbirth services.

We provide evidence on possible mechanisms through which institutional quality affects cesarean section rates. Among them, we find that higher institutional quality maps into pro-active policies (i.e. hospitals’ payment policies) to reduce inappropriate provision, which in turn lead to lower cesarean section rates. We also show that institutional quality does not lead to worse mothers’ health outcomes as measured by emergency readmissions. Finally, we investigate if institutional quality affects not only appropriateness of care, as measured by cesarean section rates, but also clinical quality. We show that higher institutional quality reduces heart attack mortality rates, but does not affect stroke mortality or mortality for chronic obstructive pulmonary disease.⁴ Overall, our results suggest that a worse institutional environment weakens hospitals’ incentives to select treatments that are cost-effective.

Our study brings together two strands of the literature. First, we contribute to the literature on the effect of the quality of institutional environment in the public sector (e.g., Mauro, 1998, Nannicini et al., 2013, Castro et al., 2014, Wong et al., 2017) by highlighting the importance of the health sector. Second, we contribute to the literature on the causes of inappropriateness in healthcare provision (e.g., Brown, 1996; Dubay et al., 1999, Gruber et al., 1999, Currie and MacLeod, 2008, Epstein and Nicholson, 2009, Francese et al., 2014, Johnson and Rehavi, 2016, Foo et al., 2017) by highlighting the role of institutional quality in affecting physician behavior.

Only very few studies have investigated the role of institutional quality in the healthcare sector. Di Tella and Schargrodsky (2003) analyze the prices paid by public hospitals for basic inputs during a crackdown on corruption in Buenos Aires, finding that prices decreased by about 15% relative to the pre-crackdown period. Azfar and Gurgur (2008) study healthcare provision in the Philippines, and report that a higher corruption in municipalities decreases immunization rates, delays vaccination of newborns and increases waiting time. Cavalieri et al. (2017) study the execution of public contracts for infrastructure in the hospital sector, and show that the performance in the provision of infrastructure is negatively affected by environmental corruption. With respect to this literature, we focus on inappropriate behavior in healthcare provision rather than prices or the execution of infrastructures in the hospital sector. Moreover, we also show that, while corruption is among the most relevant environmental factors, it is not the only source of waste in healthcare.

The study, which is closest to ours, is Francese et al. (2014). It investigates the impact of institutional features, such as supply and pricing policies, and some political economy indicators (namely, occupation and years of experience of the regional president), on regional cesarean section rates in Italy over the period 1998–2005. In their regional level analysis, they include regional fixed effects and thus exploit variation over time to identify the determinants of cesarean sections. In our study, we investigate a range of dimensions of local institutional quality in which hospitals operate. As shown below, such institutional quality varies little over time, while it varies significantly across regions.⁵ We therefore exploit variation across regions and employ an instrumental variable strategy to address the potential omitted-variable bias. Our unit of analysis is at the more disaggregated hospital level rather than regional level (though institutional quality varies across regions), and we use the more recent data for the period 2007–2012.

The rest of the study is organized as follows. Section 2 describes the Italian healthcare system. Section 3 provides a conceptual framework which discusses the mechanisms through which local institutional quality may affect the appropriateness of healthcare provision. Section 4 describes the data. Section 5 provides the empirical strategy. Section 6 gives the results. Section 7 concludes and draws implications for healthcare policy.